LFM (Liquid Foundation Models)
LFM (Liquid Foundation Models) is a model family from Liquid AI13, with 3 assessed releases in this catalog. Licenses and availability are recorded for each release, not for the family; its releases use the LFM Open License v1.0.
LFM is Liquid AI's series of hybrid language and multimodal models for on-device use, built from gated short-convolution blocks and a small number of grouped-query attention blocks. LFM2 (from July 2025) introduced open-weight dense models from 350M parameters upward and later an 8.3B mixture-of-experts model. LFM2.5 (from January 2026) extended pretraining and scaled up reinforcement-learning post-training; 2026 releases include LFM2.5-8B-A1B (May), LFM2.5-2.6B (August), and LFM2.5-VL-3B (August), alongside vision, audio, and encoder variants.43516710
- Website: Liquid AI models page (external site: liquid.ai)
- Documentation: LFM documentation (external site: docs.liquid.ai)
- Paper: LFM2 technical report (arXiv 2511.23404) (external site: arxiv.org)
- License: LFM Open License (external site: liquid.ai)
On AI on Hugging Face: LiquidAI/LFM2.5-VL-3B
Releases assessed
| Release | Weights | Tier (USASI rubric v0.2) | License | Released |
|---|---|---|---|---|
| LFM2.5-2.6B | Weights: Public | Model-disclosure tier (USASI rubric v0.2): Open-weight | LFM Open License v1.0 | Aug 4, 2026 |
| LFM2.5-24B-A2B | Weights: Public | Model-disclosure tier (USASI rubric v0.2): Open-weight | LFM Open License v1.0 | Oct 13, 2026 |
| LFM2.5-8B-A1B | Weights: Public | Model-disclosure tier (USASI rubric v0.2): Open-weight | LFM Open License v1.0 | May 28, 2026 |
What it is useful for
Liquid AI positions the models for on-device and other resource-constrained deployment, and its model cards recommend the current text models for agentic workflows, tool use, data extraction, retrieval-augmented generation, and structured outputs rather than knowledge-heavy or heavy coding tasks.289
Organization context
Provenance and derivatives
The LFM language models are pretrained by Liquid AI itself; the current text instruction and reasoning models are post-trained from Liquid's own base checkpoints (for example LFM2.5-2.6B-Base and LFM2.5-8B-A1B-Base). The vision-language models are not text-only LFMs: the LFM2.5-VL-3B model card says it combines the LFM2.5-2.6B language backbone with a SigLIP2 NaFlex vision encoder. Liquid's LFM2 announcement says an earlier Liquid model, LFM1-7B, was used as a teacher for knowledge distillation during LFM2 training.89114
In the news
Liquid AI releases LFM2.5-24B-A2B, an open-weight mixture-of-experts model for on-device agents
Liquid AI announced LFM2.5-24B-A2B on October 13, 2026, describing it as a reasoning model built for fast, private agents on-device and as the largest and final model in the LFM2.5 family. The release includes a base model (LFM2.5-24B-A2B-Base) and a post-trained model (LFM2.5-24B-A2B), published as open weights on Hugging Face. The model card lists 24.0B total parameters with 2.45B active per token and a context length of 131,072 tokens, and says the model is not recommended for knowledge-heavy tasks on its own. The weights are released under the LFM Open License v1.0. USASI has not tested this model.
Liquid AI releases open-weight d1 decision models for text, vision, and audio
On October 7, 2026 Liquid AI announced "Open d1", two open-weight decision models that return an answer in a single forward pass instead of generating text. The announcement and model cards describe d1-3B, built on LFM2.5-VL-3B, which takes text and images, and d1-omni-600M, built on LFM2.5-Encoder-350M, which takes text with images or text with up to 30 seconds of speech. The weights are published on Hugging Face, and the announcement says the models have day-one support in llama.cpp. Both model cards give the license as "other" with the name lfm1.0, pointing to a LICENSE file in each repository; the announcement itself does not name a license.
Liquid AI releases LFM2.5-VL-3B-DSpark, a draft model for faster vision-language decoding
Liquid AI announced LFM2.5-VL-3B-DSpark on September 24, 2026. It is an experimental draft model of about 280 million parameters that speeds up text generation by its LFM2.5-VL-3B vision-language model through speculative decoding; Liquid AI says the target model's output is unchanged. The weights are on Hugging Face in Safetensors and GGUF formats, with support in SGLang, MLX-VLM, and llama.cpp. They are released under the LFM Open License v1.0, the same license as other LFM2.5 releases, which limits commercial use by larger companies through an annual-revenue threshold.
Liquid AI releases LFM2.5-VL-3B, an open-weight vision-language model for on-device use
Liquid AI announced LFM2.5-VL-3B on August 12, 2026, a non-reasoning vision-language model that it says answers directly to keep latency low for real-time and on-device applications. The model card says it uses the LFM2.5-2.6B language model as its backbone with a SigLIP2 NaFlex vision encoder, and lists a context length of 32,768 tokens. Weights are on Hugging Face with GGUF, ONNX, and MLX exports, under the LFM Open License v1.0. USASI has not tested this model.
Liquid AI releases LFM2.5-2.6B, an open-weight on-device model trained for agentic workloads
Liquid AI announced LFM2.5-2.6B on August 4, 2026, describing it as a 2.6B-parameter model trained for agentic workloads such as planning, calling tools, and multi-step tasks on devices including phones and CPUs. The base model (LFM2.5-2.6B-Base) and the post-trained model (LFM2.5-2.6B) were published on Hugging Face. The model card lists a context length of 131,072 tokens and says the model is not recommended for agentic coding and knowledge-heavy tasks. The weights are released under the LFM Open License v1.0. USASI has not tested this model.
U.S. eligibility
Sources
This listing is not an endorsement, a safety assessment, or a federal approval.
Share or cite this page
- X (opens X)
- Bluesky (opens Bluesky)
- LinkedIn (opens LinkedIn)
- Reddit (opens Reddit)
- Hacker News (opens Hacker News)
- Email (opens your email app)
- RSS updates
Cite this page
APA style
USASI. (2026, September 29). LFM (Liquid Foundation Models). United States of America Superintelligence. https://unitedstatesofamericasuperintelligence.com/open/lfm/
BibTeX
@misc{usasi_lfm,
author = {{USASI}},
title = {LFM (Liquid Foundation Models)},
year = {2026},
month = sep,
howpublished = {\url{https://unitedstatesofamericasuperintelligence.com/open/lfm/}},
note = {United States of America Superintelligence. Last updated 2026-09-29}
}The date is when this page was last updated. Add the date you read it if your style needs one.
For maintainers: a “Listed in USASI” badge
Links to this page. It says the entry is listed, not endorsed or rated. About the badge
Markdown (for a README)
[](https://unitedstatesofamericasuperintelligence.com/open/lfm/)
HTML
<a href="https://unitedstatesofamericasuperintelligence.com/open/lfm/"><img src="https://unitedstatesofamericasuperintelligence.com/badges/listed.svg" alt="Listed in the USASI catalog" height="20"></a>